Key insights:
Rudy Arora and Sarthak Dhawan are 21 years old. They make over a million dollars a month. Their app, Turbo AI, has 8 million users. But the story didn't start with AI. It started with Christmas lights, stolen cookies, and a middle school friendship in Texas.
Sarthak started coding in third grade. By ninth grade, he built an app called Gradeify that pulled 40,000 daily active users from word of mouth alone. He would walk around school and see classmates using his app without knowing he made it. That feeling hooked him on building things.
Around the same time, Rudy noticed his family struggled to find contractors for Christmas lights. He built a Google Form, knocked on doors, and collected orders. Then he realized an app would work better. He knew Sarthak was technical, so they partnered up and created WorkBe, a marketplace for Christmas light installation. Over two and a half years, they made $60,000 in revenue with roughly 10% margins.
The margins were terrible. One order might earn them $40 after insane amounts of effort. Logistics nightmares, unreliable contractors, and angry customers made them rethink everything. Sarthak once had to buy a string of Christmas lights off Amazon and personally deliver them to a customer whose contractor ghosted everyone.
The math was simple. With WorkBe, they earned $40 per order with no guarantee of repeat business. With software, someone signs up online, pays by themselves, and you get money on a recurring basis with high margins. Every single advantage pointed toward software.
Around the same time, AI was taking off. ChatGPT had just launched. Consumer AI apps barely existed. They saw an opportunity to sell AI tools on a recurring subscription basis. There were almost no AI study tools on the market when they started ideating.
The idea came from a personal problem. Sarthak hated making flashcards for AP US History. He would paste content into GPT, export it, and manually create study materials. Both founders hated listening to professors talk. What if AI could take notes, make flashcards, and generate quiz questions automatically? That became TurboLearn.
The MVP took about two to three months to build. This was before AI coding tools existed. GPT 3.5 was out and, in Sarthak's words, "borderline useless when it comes to anything technical." Today, he says the same MVP would take one to two days.
The initial product let you upload a PDF and used the GPT API to convert it into notes, flashcards, and quizzes. The quality was, by their own admission, "pretty abysmal." It had bugs. It didn't work very well. But it existed, and that was enough to start testing.
Getting users was the hardest part. Before they cracked viral marketing, the founders tried everything from bathroom posters to bribing students with cookies. Their journey from zero to 8 million users holds lessons for anyone building a product today.
Rudy would steal cookies from the dining hall at Northwestern, set up a booth, and tell students, "Sign up for Turbo and I'll give you a cookie." He gave out roughly a thousand cookies. Sarthak did the same thing at Duke. People got so annoyed that they started calling Sarthak "Turbo" on campus.
They also printed posters with a poop emoji and the tagline "Shitty professor who talks too fast?" and stuck them inside every bathroom stall at Duke. It was hilarious but not effective. The funny part is that at least 20 other student startups at Duke copied the bathroom poster strategy, thinking that was how Turbo grew to 8 million users. It wasn't.
These scrappy tactics taught them something important. Attention without conversion is worthless. They needed a system that could scale.
The real unlock was UGC creators. They take college students, have them start brand new social media pages, and every post on that page promotes Turbo AI. But here is the key difference. If you see an account called "TurboLearn AI" posting a video, you know it is an ad. If you see "Sarah's spam account" posting about Turbo, you think it is a real person recommending a product.
They now manage over 500 creators who post every single day. Seven full-time creator managers oversee these relationships. Creators get a flat rate per video plus incentives. Their best creator earned almost $30,000 in a single month and is now a full-time employee.
The secret to retaining creators? Be their friend. When competitors offered creators over $15,000 monthly retainers, many chose to stay with Turbo because of the relationships they had built with the team.
Their CMO Nick noticed that the most viral influencers on TikTok always do something physical while talking. They shave, make smoothies, or cook. So Turbo started having creators cut fruit while promoting the app. It looked like a "get ready with me" storytime video, not an ad.
They would Venmo creators $20 a week to buy fruit and props. For two to three months, every Turbo creator was cutting fruit. It defined their brand. The views and conversions picked up dramatically because viewers thought they were watching a real influencer, not a company account.
Eventually, fruit became so associated with Turbo ads that viewers started commenting, "Not another Turbo ad." So they evolved. Now creators pour Kool-Aid into water, stir drinks, or use other props. The fundamental principle stays the same: do something engaging in the video so it feels like real content, not advertising.
The founders shared a detailed framework for anyone wanting to build an app that makes money. Their advice is surprisingly grounded and goes against a lot of the hype you see online about building apps in five minutes.
The founders broke it down into two critical steps:
A lot of apps fail because they nail attention but cannot monetize. Social apps where you "connect with friends" rarely get people to pay. The founders stressed that both virality and willingness to pay must exist before you invest serious time building.
The founders gave an honest answer that most people do not want to hear. If you have zero coding experience, $100, and no technical help, you probably should not build an app. The problem is you do not know what you do not know. When something breaks, you cannot fix it. You cannot even identify why things are going wrong.
Their recommended approach for non-technical people is to use Claude Code and ask an enormous number of questions. Start with "How are apps built?" and work your way up. Every time you see a word you do not understand, ask what it means. Within a month, you can get a foundational understanding and something on the app store.
The important nuance is that AI coding tools work best for things that have been built thousands of times before. A tic-tac-toe app? AI will one-shot it. But businesses make money because they do something new. If the LLM has seen your idea a million times, it means the idea is easy, and if it is easy, someone has already done it. There is a fundamental tension between what AI can build easily and what is actually worth building.
The founders were refreshingly honest about the economics of consumer apps. When you see headlines about 18-year-olds making $30 million a year, the asterisk is that marketing spend behind that growth is enormous. The actual cash flow, how much the bank account increases each month, is a vastly different story.
AI businesses face an additional challenge. Unlike traditional software where maintaining a user costs almost nothing, LLM inference is expensive. They gave the example of Cursor, the AI coding tool. Because Cursor must use the best models to keep developers happy, their gross margins on paying subscribers are not great. A subscriber paying $20 per month might cost Cursor more than $20 in AI model usage.
Their own engineering team of 10 people spends $20,000 to $30,000 per month on Claude Code alone. The unit economics of AI businesses are fundamentally different from traditional software, and most revenue headlines are misleading.
With hundreds of millions of views across social media, the Turbo AI founders have learned hard lessons about the difference between viral content and content that actually drives users and revenue.
This is what the founders called "the hardest question of all time." They had a two-month stretch where they racked up 600 to 700 million views on social media. User growth barely changed from the previous month. The insight was brutal: a video that goes viral for the wrong reason does not convert.
The reason your video goes viral has to be because of the product, not despite the product. Your product must be the solution to a problem that the video builds up. If someone makes a video about basketball for 30 seconds and then randomly plugs a study app, viewers feel tricked. The content and the product are disconnected.
A bad example: recording your mom yelling at your dad and turning it into notes with Turbo. The video goes viral because of the drama, not because of the product. A good example: the UMAX integration with looks-maxing creator K Shami. He did his standard "Mogwarts" format showing how to improve someone's appearance, and simply included a screenshot of the UMAX app as the tool the person used. The product was directly the solution to the problem in the video.
Most companies look for viral ideas within their own niche. The Turbo founders do the opposite. They pull ideas from completely unrelated niches. Rudy maintains five different TikTok "For You" pages on his phone, each tuned to a radically different audience. One is what he calls the "depressed white girl crying over her ex" feed.
Their most creative campaign came from the manifestation niche on TikTok, where creators light candles and tell viewers they will have a great year. Turbo flipped this into creators manifesting good grades for viewers, with Turbo AI as the solution. Creators went into their rooms, turned LED lights purple, and spoke in a psychic tone about how Turbo would help them crush the semester. It generated massive views and decent conversions.
The lesson is clear. If you only look at what competitors in your space are doing, you will always be derivative. The best ideas come from studying what works in spaces that have nothing to do with your product and finding creative bridges.
With tens of thousands of videos active at any given time, the founders admitted that tracking where users come from is essentially an unsolvable problem. Someone might see a TikTok video, not click the link, hear about the app from a friend the next day, and then search for it organically. There is no way to connect that signup to the original video.
They shared a fascinating anecdote. Someone running a fund responsible for over a billion dollars of Grammarly's advertising spend told them that even at that scale, a lot of marketing decisions are "vibes based." Companies spending $50 million on advertising are surprisingly not data-driven about attribution. That should give every founder some reassurance that nobody truly knows what they are doing.
Their advice for early-stage founders: when you only have 50 users a day and a video gets 5 million views, you can clearly see the spike. Use that early stage to experiment and roughly attribute which content formats drive growth. Once you scale to thousands of videos and tens of thousands of daily signups, precise attribution becomes pointless.
Both founders dropped out of elite universities. Sarthak left Duke. Rudy left Northwestern. Their perspectives on education, hiring, and what drives success are shaped by building a company worth over $100 million before turning 22.
The founders had nuanced but different takes. Rudy argued that if you are going to college purely as a financial investment to become more employable, the ROI does not make sense unless you are on scholarship or at an affordable state school. Trade school can produce the same salary.
Sarthak disagreed. He argued that because AI makes it easy for anyone to learn anything, the noise in job application pools has become overwhelming. Employers need filters. The first filter any company uses is which college you attended. With AI reducing the number of available jobs, these filters become even more important.
Their own hiring proves the point. When they have 10,000 applicants and time to talk to 10, they filter by Ivy League attendance. Not because graduating from Harvard means anything, but because getting into Harvard signals something about the person. They actually prefer Ivy League dropouts over graduates, because the dropout demonstrated both the ability to get in and the courage to leave.
Sarthak does not ask engineering candidates what programming languages they know. He does not care about their coursework. The only thing he looks for is a history of excellence in anything. A national champion rower. A top 100 League of Legends player. Someone who was the best in their school at debate.
His reasoning: the skills required to be top 0.1% in any discipline are remarkably similar. Waking up day after day, putting time into one thing, continuously improving, and doing it over a very long period. If someone has demonstrated that in any area, those skills transfer. If they cannot point to something they are genuinely excellent at, it is an instant negative signal.
Sarthak's answer was counterintuitive. In a world where you can learn anything instantly, the most valuable skill is learning to not learn a million things. Focus on one thing and double down. The frameworks for success are simple. Want to get better at coding? Code more. Want to get better at running? Run more. Simple does not mean easy.
AI has created massive shiny object syndrome. Every week there is a new tool, a new opportunity, a new skill to learn. The people who win are the ones who resist the urge to chase everything and instead compound their efforts in one direction over a long period of time.
The founders closed with advice that applies to anyone building something. Sarthak emphasized biasing towards action. The gap between thinking about something and actually doing it is where most people lose. Thinking feels productive but accomplishes nothing. If you want to go for a run, go now. If you have an idea, test it today. The compounding effects of acting quickly are enormous.
Rudy shared advice from a mentor: the skills that take you from zero to one are radically different from the skills that take you from one to ten. Managing people, empowering teams, staying organized, these are not the same skills as scrapping together an MVP and getting your first users. Recognizing that shift early saves you months of struggle.
If you are someone who wants to build a career around automation and AI rather than be replaced by it, the Complete RPA Bootcamp is worth looking into. It takes you from beginner to professional in Robotic Process Automation, Agentic Automation, and Enterprise Orchestration. Instead of watching AI take jobs, you become the person building the automation. You can apply at apply.completerpabootcamp.com.
For the full conversation, including live demos of Turbo AI, stories about bathroom poster marketing, and the founders debating whether college is worth it, watch the complete interview embedded below on the Jack Neel YouTube channel. It is over two hours of practical insights from two of the most successful young founders in the world right now.
